OK, I decided to try to get some data. I enabled full DBus debugging [1] and started the Terminal app with dbus-monitor --system running.
I walked from my 3rd-floor apartment to the front door, with Terminal open, watching top in a second tab. The attached log (dmon2.log) is what resulted. dbus-daemon spiked around 90% as I got to the bottom. Then settled. So not the constant spiking that we've seen in this bug. But something.
I will also attach a second file (dmon3.log) that is me walking from my front door down the street, to be away from any vestige of my wifi. dbus-daemon spiked again for a solid chunk of time. Eventually it finished, around the same time that the network-indicator finally updated to notice that we weren't on wifi anymore.
Unfortunately, dbus-daemon doesn't do timestamps by default. So I made another walk out and back again with the --profile argument. Which isn't as verbose, but provides timestamps. That's attached as dmon5.log.
I haven't done much analysis of these yet. But looks like a lot of cpu-device-online signals from upstart, wireless properties changing from NetworkManager, and attempts to upload apport crash files. Will dig into it, but wanted to provide raw data first.
OK, I decided to try to get some data. I enabled full DBus debugging [1] and started the Terminal app with dbus-monitor --system running.
I walked from my 3rd-floor apartment to the front door, with Terminal open, watching top in a second tab. The attached log (dmon2.log) is what resulted. dbus-daemon spiked around 90% as I got to the bottom. Then settled. So not the constant spiking that we've seen in this bug. But something.
I will also attach a second file (dmon3.log) that is me walking from my front door down the street, to be away from any vestige of my wifi. dbus-daemon spiked again for a solid chunk of time. Eventually it finished, around the same time that the network-indicator finally updated to notice that we weren't on wifi anymore.
Unfortunately, dbus-daemon doesn't do timestamps by default. So I made another walk out and back again with the --profile argument. Which isn't as verbose, but provides timestamps. That's attached as dmon5.log.
I haven't done much analysis of these yet. But looks like a lot of cpu-device-online signals from upstart, wireless properties changing from NetworkManager, and attempts to upload apport crash files. Will dig into it, but wanted to provide raw data first.
[1] https:/ /wiki.ubuntu. com/DebuggingDB us